The Reason I Keep Buying Amazon That Wall Street Keeps Missing

AWS backlog was cited at $364B (Q1 2026 commercial backlog) and AWS grew 28% YoY to $37.59B with a 37.7% operating margin, supporting the thesis that capex is backed by contracted demand (including an AI-related Anthropic deal for over $100B). The article also flags AI silicon traction with a ~$20B annual revenue run rate and Trainium commitments over $225B, while advertising crossed $70B in TTM revenue (+24%) and consolidated operating margin hit 13.1% (all-time high). The primary risk highlighted is cash: FCF TTM fell 95% to $1.2B on $44.2B capex in a quarter and long-term debt rose to $119.1B; however, management commentary claims customer commitments cover much of the 2026 spend. Overall, the stock is described as trading around $247.49 versus a $312.91 consensus target with 62 buy ratings and no sells.

Analysis

The market is still framing this as a FCF problem, but the more important mechanism is balance-sheet timing: Amazon is front-loading depreciation while deferring the revenue realization on a largely pre-sold capacity base. That creates a near-term cash squeeze and headline risk, but if utilization ramps as planned, the spend should behave more like a financed asset refresh than speculative capex. Over the next 1-3 months, the stock will trade on whether management can keep convincing investors that incremental capex is matched by contracted demand, not on the backlog figure itself.

Relative winners are the companies with exposure to AWS monetization breadth and ad-ops leverage: AMZN itself, and to a lesser extent suppliers into data-center power/networking. Relative losers are the hyperscalers that still need external AI ecosystems to prove differentiation; MSFT is more exposed to a perception reset if OpenAI becomes more multi-homed, while GOOGL remains the third-choice cloud story unless it can show a sharper AI monetization path. A subtler loser is NVDA on AWS-specific workload mix, because custom silicon reduces merchant GPU intensity even if overall AI capex stays elevated.

The contrarian miss is that commitments are not cash: a backlog does not protect the stock if the conversion period stretches or utilization disappoints. The key falsifier is sustained free cash flow near zero into the next two quarters while debt continues rising and AWS growth slips below the mid-20s; in that scenario the equity will start trading like a capital-intensive utility instead of a compounder. If Trainium savings show up in operating margin before then, the current multiple is probably too low.

Key Decisions for Investors

• Add AMZN on weakness / buy 6-month 250/300 call spreads into any dip below 245; risk is the stock stays range-bound if FCF does not inflect, reward is multiple expansion if AWS growth and margin stay strong.
• Pair trade: long AMZN / short MSFT over 3-6 months to express the view that AWS has the cleaner AI monetization loop while MSFT carries more roadmap risk from OpenAI dependency; cover if Azure growth re-accelerates or MSFT narrows the valuation gap.
• Smaller RV expression: long AMZN / short GOOGL over 1-3 months if you want exposure to cloud-share and AI-lab commitment momentum; thesis breaks if GOOGL shows materially better cloud margin leverage or ad growth surprise.
• Use TGT as a consumer/retail hedge against AMZN strength: short TGT against long AMZN on any retail-segment weakness in coming prints; falsify the trade if Amazon retail growth slows materially or TGT shows share stabilization.
• Set a hard watch item on AMZN free cash flow and net debt over the next two quarters; if FCF remains near zero while capex stays elevated, reduce exposure regardless of backlog rhetoric.